Tuning the support adsorption properties of Pd/SiO2 by silylation to improve the selective hydrogenation of aromatic ketones

Article abstract of DOI:10.1016/j.jcat.2008.04.006

Silylation of Pd/SiO₂ catalysts increases the selectivity toward alcohols in the reduction of aromatic ketones. This work demonstrates that the selectivity is directly related to the adsorption strength of the alcohol on the surface of the support relative to the adsorption strength of the ketone. This observation can be explained by interaction of the support coverage with the metal coverage. Silylation yields a more hydrophobic support, on which the aromatic alcohol adsorbs more weakly relative to the ketone, in turn decreasing the amount of the alcohol adsorbed on the metal and thus suppressing the consecutive reduction of the alcohol. Silylation was carried out by using di-alkyl (dichlorodimethylsilane) and tri-alkyl (hexamethyldisilazane and hexamethyldisilane) silylating agents. Hexamethyldisilazane provided to be the most effective agent in terms of incorporation of methyl groups, catalyst hydrophobicity, and stability. Selective hydrogenation of 4-isobutyl acetophenone (4-IBAP) to 4-isobutylphenyl ethanol (4-IBPE) revealed that not only was the fresh hexamethyldisilazane-silylated Pd/SiO₂ catalyst more selective than the untreated catalyst, but also the silylated catalyst was much more selective after a deactivation-regeneration cycle than the untreated Pd/SiO₂ catalyst. The change in selectivity can be explained by a change in the relative adsorption strength of 4-IBPE over 4-IBAP on silylation.

Full text of DOI:10.1016/j.jcat.2008.04.006

Journal of Catalysis 257 (2008) 55–63
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Tuning the support adsorption properties of Pd/SiO by silylation to improve
2
the selective hydrogenation of aromatic ketones
∗
A. Quintanilla , J.J.W. Bakker, M.T. Kreutzer, J.A. Moulijn, F. Kapteijn
Catalysis Engineering, DelftChemTech, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands
a r t i c l e i n f o
a b s t r a c t
Article history:
Silylation of Pd/SiO2 catalysts increases the selectivity toward alcohols in the reduction of aromatic
ketones. This work demonstrates that the selectivity is directly related to the adsorption strength of the
alcohol on the surface of the support relative to the adsorption strength of the ketone. This observation
can be explained by interaction of the support coverage with the metal coverage. Silylation yields a
more hydrophobic support, on which the aromatic alcohol adsorbs more weakly relative to the ketone, in
turn decreasing the amount of the alcohol adsorbed on the metal and thus suppressing the consecutive
reduction of the alcohol. Silylation was carried out by using di-alkyl (dichlorodimethylsilane) and tri-
alkyl (hexamethyldisilazane and hexamethyldisilane) silylating agents. Hexamethyldisilazane provided to
be the most effective agent in terms of incorporation of methyl groups, catalyst hydrophobicity, and
stability. Selective hydrogenation of 4-isobutyl acetophenone (4-IBAP) to 4-isobutylphenyl ethanol (4-
IBPE) revealed that not only was the fresh hexamethyldisilazane-silylated Pd/SiO2 catalyst more selective
than the untreated catalyst, but also the silylated catalyst was much more selective after a deactivation–
regeneration cycle than the untreated Pd/SiO2 catalyst. The change in selectivity can be explained by a
change in the relative adsorption strength of 4-IBPE over 4-IBAP on silylation.
Received 21 February 2008
Revised 8 April 2008
Accepted 9 April 2008
Available online 21 May 2008
Keywords:
Pd catalyst
Silylation
Physisorption
Selective hydrogenation
Hydrophobicity
©
2008 Elsevier Inc. All rights reserved.
1
. Introduction
Previous studies have reported higher activity, selectivity, and
stability of the silylated catalysts compared with their untreated
counterparts. Corma et al. [1] correlated selectivity with the per-
centage of the surface coverage by the methyl groups incorporated
on the catalyst. The improved performance is attributed to the
hydrophobic properties of the catalyst surface, which particularly
affects (i.e., weakens) the adsorption of the oxygenated products;
however, the control experiments, in which the relative adsorption
strength of the various reactant and products were compared, were
not reported.
In the present work, we focused on selective hydrogenation
of aromatic ketones to aromatic alcohols [14]. We investigated
the industrially important hydrogenation of 4-isobutyl acetophe-
none (4-IBAP) to 4-isobutylphenyl ethanol (4-IBPE) using sup-
ported metal catalysts. This reaction is an intermediate step in
the catalytic route developed for ibuprofen, a well-known anal-
gesic [15]. Pd/SiO2 catalysts are particularly active and selective
when hydrogenation of 4-IBAP is carried out in a nonpolar sol-
vent, such as n-decane [16]. The hydrogenation route is shown in
Fig. 1 [17]. The following kinetic mechanism has been proposed for
this reaction [18]:
An appropriate modification of the hydrophobic–hydrophilic
properties of catalyst can optimize the relative adsorption
a
strengths of the reactant and products on a support, and thus
the reaction selectivity as well. The most common strategy for in-
creasing hydrophobicity is silylation of the catalyst by alkyl- and
chloroalkylsilanes in the gas or liquid phase [1–9]. Silylation re-
duces the silanol groups of the support, thereby decreasing the
adsorption of water and other polar products on the catalyst sur-
face. Hexamethyldisilazane (HMDS) is the most widely used sily-
lating agent due to its high reactivity with silica [7,10].
Silylation has been performed mainly with mesoporous ma-
terials, typically Ti-MCM-41 and 48, because of their instability
in presence of water. The reactions in which silylated catalysts
were tested were mainly selective oxidations, including epoxida-
tion [1–3,8–11], hydroxylation of benzene [11], and oxidative desul-
furization of diesel [12], in which the oxidized product had a
strong affinity to the catalyst. These catalysts also have been used
in the synthesis of alkylamines [13], isomerization reactions with
HZMS-5 [6] and Fischer–Tropsch synthesis [7].
IBAP + * ↔ IBAP*
H2 + * ↔ 2H*
IBAP* + H* ↔ IBAP–H* + *
(1)
(2)
(3)
Corresponding author. Fax: +31 15 27 85006.
E-mail address: m.a.quintanillagomez@tudelft.nl (A. Quintanilla).
*
0
021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.04.006

56
A. Quintanilla et al. / Journal of Catalysis 257 (2008) 55–63
Fig. 1. Reaction scheme for 4-IBAP hydrogenation with Pd/SiO2.
IBAP–H* + H* → IBPE* + *
IBPE* ↔ IBPE + *
IBPE* + H* ↔ IBPE − H* + *
IBPE–H* + H* → IBEB + H2O* + *
H2O* ↔ H2O + *
Here, Eqs. (4) and (7) are the rate-determining steps, and the
others are adsorption equilibria on the palladium sites, indicated
by *. In this work, we also are interested in the support–metal
interaction of adsorbed species. Denoting a physisorption site on
the support by #, for IBAP, we write
(4)
(5)
(6)
(7)
(8)
673 K at a rate of 1 K/min and calcined in static air at 673 K for
4
h.
2
.1.1. Silylation
Silylation reactions were carried out in the liquid phase with
the catalyst. Only in the case of DDMS was silylation carried out
with the silica support before Pd application, to avoid leaching
of Pd induced by the presence of chloride. In all cases, the con-
centration of the silylating agent was 5% by volume, which was
sufficiently high according to the study of Cagnoli et al. [3], in
which five different concentrations of HMDS in toluene, ranging
from 0.8% to 16% in volume, were tested. The operating conditions
and the chemical reactions for each silylating agent are depicted in
Fig. 2. The silylated catalysts were filtered, washed with ethanol,
and dried at 383 K in air for 4 h. Hydrophobic silica also was
prepared by silylation with HMDS, to study the interaction of the
support only with the reactants and products.
IBAP + # ↔ IBAP#
IBAP# + * ↔ IBAP* + #
(9)
(10)
where Eq. (9) represents physisorption on the support and Eq. (10)
demonstrates how strong adsorption on the support enhances the
amount adsorbed on the metal by, for example, surface diffu-
sion [19]. Analogous to these equations for IBAP, similar expres-
sions can be written for IBPE, IBEB, and H2O.
Continuous flow operation of this reaction currently is not pos-
sible, because the catalyst deactivates with loss of selectivity. Re-
generation of the catalyst after reaction, involving heat treatment
in a static air oven at 623 K for 4 h, is required. This treatment
allows recovery of the activity and, more importantly, makes the
catalyst no longer selective toward alcohol.
The aim of this work was to demonstrate that silylation of
Pd/SiO2 catalyst also can increase selectivity in hydrogenation.
We also found that, more remarkably, the selectivity was main-
tained after regeneration. We studied the silylation of Pd/SiO2
catalyst using silylating agents with different functional groups:
hexamethyldisilazane [(H3C)6Si2NH, HMDS], dichlorodimethylsi-
lane [(H3C)2Si(Cl)2, DDMS], and hexamethyldisilane [(H3C)6Si2,
HMS]. HMDS and DDMS have been reported to be effective sur-
face modification agents for increasing the hydrophobicity of SiO2
2.2. Catalyst characterization
Fresh catalysts “as prepared” and spent catalysts after wash-
ing with ethanol and dried at 383 K for 2 h at a heating rate
of 0.5 K/min were analyzed by several techniques. Transmission
electron microscopy (TEM) was performed using a Philips CM30UT
electron microscope with a FEG as a source of electrons operated
at 300 kV. Samples were mounted on Quantifoil microgrid carbon
polymer supported on a copper grid by placing a few droplets of
a suspension of ground sample in ethanol on the grid, followed by
drying at ambient conditions.
The catalysts were analyzed by inductively coupled plasma op-
tical emission spectroscopy (ICP-OES) performed in a Perkin–Elmer
Plasma 2000 apparatus to determine the palladium content. For
the analysis, ca. 35 mg of the sample was digested in 50 mL of an
acid mixture (0.7% HCl, 1% HF, and 1.25% H SO ).
2
4
Thermogravimetric analysis (TGA) of the catalysts was carried
e
out on a Mettler–Toledo TGA/SDTA851 thermobalance. The sample
powders were heated in air from 298 to 1173 K at heating rate of
10 K/min.
[20,21]. HMS has been successfully used in the silylation of allylic
acetates in the presence of Pd catalysts [22]. We compared the ac-
tivity, selectivity, and stability of the different Pd/SiO2 catalysts and
correlated improvements in selectivity with the relative adsorption
strength of 4-IBPE to 4-IBAP.
Diffuse reflectance infrared Fourier transform (DRIFT) spectra
were recorded with a single beam Nicolet Magna 550 FTIR spec-
trometer in a Spectratech DRIFTS accessory, against a background
recorded for a sample cup filled with KBr. The DRIFTS chamber
was flushed with N2. The spectra were obtained by collecting 256
−1
2. Experimental
scans at 8 cm
units.
resolution and displayed in absorption reflectance
2
.1. Catalyst preparation
Temperature-programmed oxidation–mass spectrometry (TPO-
MS) measurements were performed in a conventional flow appa-
ratus using a quartz microreactor, with 0.1 g of the sample heated
at a rate of 10 K/min from 298 to 1173 K in 20% O2 in He (flow
We prepared 2.5 wt% Pd/SiO2 catalysts through the wet im-
pregnation method, using Pd acetate trimer (Alfa Aesar) as the
palladium precursor and Davisil 643 (35–74 μm) as the SiO2 sup-
port [16]. To prepare the catalyst, the required amount of precursor
solution was dissolved in an excess of toluene and added to the re-
quired amount SiO2 support suspended in toluene. The mixing was
carried out overnight under continuous stirring. The catalyst was
filtered under vacuum, dried at 373 K for 16 h, and then heated to
3
rate of 50 cm /min). The gaseous species from the sample were
monitored by a quadrupole mass spectrometer (Balzers, PRISMA)
connected online with the reactor. Mass spectra were recorded in
multiple-ion detection (MID) mode using a Channeltron detector.
The porous structure of the catalysts was characterized from the
77 K N2 adsorption/desorption isotherms using a Quantachrome

A. Quintanilla et al. / Journal of Catalysis 257 (2008) 55–63
57
Fig. 2. Operating conditions of the silylation reactions of Pd/SiO2 catalyst with different silylating agents.
Autosorb-6B apparatus. The samples were first pretreated in situ by
2.3. Selective hydrogenation
drying in vacuum at 403 K to remove moisture and other volatile
compounds. The isotherms provided information on the specific
surface area (SBET), pore volume (VP), and pore diameter (dPore).
Hydrogenation reactions were performed in a stainless steel
reactor equipped with gas-induced stirrer (Premex Reactor
a

58
A. Quintanilla et al. / Journal of Catalysis 257 (2008) 55–63
CH-2543). In a typical experiment, after in situ reduction of catalyst
at 373 K and 10 bar of H2 for 1 h, the desired amount of reactant
Table 1
Metal loading, average Pd crystal size and textural properties of the solids
(
4-IBAP) and solvent (n-decane, containing 0.43 wt% of water, mea-
Catalyst
Pd loading
(%)
Average Pd
crystal size (nm)
S
VP
(cm /g)
dPore
(nm)
BET
2
3
sured by Karl Fischer titration) were charged to the reactor from an
external injection vessel. The contents (ca. 200 cm ) were heated
up to the reaction temperature and then pressurized with hydro-
gen to the required pressure. The progress of the reaction was
determined by monitoring the hydrogen supply rate as a function
of time. All semibatch experiments were carried out under stan-
(m /g)
3
Pd/SiO2
Fresh
In situ reduction
Spent
Regenerated
2.5
2–4
5–8
5–7
5–8
389
NM
377
326
0.81
NM
0.83
0.62
7.2
NM
7
a
7
CAT-HMDS
Fresh
Spent
dard conditions of CIBAP = 0.27 mol/L, T = 373 K, pH = 20 bar,
2
2.5
2.5
1.54
2.1
2–4
2–5
3–7
355
340
359
0.69
0.64
0.70
7
7
7.2
0
5
.1 g of catalyst, and a stirring speed of 1500 rpm (see [16]). After
h of reaction, the reactor was cooled to room temperature and
Regenerated
depressurized. Then one of the following reuse experiments was
conducted: addition of 50 mmol of 4-IBAP to the reaction mixture
through the storage vessel, to avoid exposure of the catalyst to air,
or separation of the catalyst from the reaction mixture followed by
regeneration of the catalyst in static air oven at 623 K for 4 h, to
remove the surface carbonaceous deposits. In the latter case, the
second run was performed as the fresh run, including reduction of
the regenerated catalyst.
Experiments with HMDS-silylated silica were performed un-
der the same operating conditions as for the fresh catalysts, to
study the interaction of the reactants and products with this ma-
terial under reaction conditions and to discriminate the interaction
with both Pd and the SiO2 support. For this purpose, two dif-
ferent feeds were introduced to the reactor loaded with 0.1 g of
CAT-HMDS520
Fresh
Spent
2–4
5–10
4–10
386
348
362
0.76
0.68
0.71
7
7.2
7
Regenerated
CAT-DDMS
Fresh
Spent
2–4
3–5
3–6
389
393
396
0.75
0.73
0.75
7
7
7
Regenerated
CAT-HMS
Fresh
Spent
2–4
5–8
4–7
384
NM
390
0.74
NM
0.74
7
NM
7
Regenerated
a
NM—not measured.
3
HMDS-silylated silica: ca. 200 cm of a solution of 0.27 mol/L of
4
Fig. 3 compares the differential thermogravimetric (DTG) pro-
3
-IBAP in n-decane, and ca. 200 cm of a reaction mixture contain-
files of the fresh and spent catalysts and HMDS-silylated silica. It
also indicates whether endothermal (i.e., desorption) or exother-
mal (i.e., combustion) processes occurred, as determined by the
single point differential thermal analysis (SDTA) present in the
thermobalance. The silylated fresh catalysts and support released
adsorbed water at around 330 K. The water peak was strong for
the untreated Pd/SiO2 and HMS and DDMS-silylated catalysts. At
around 830 K, the methyl groups grafted on the silica by silylation
were burnt off. For the HMDS-silylated SiO2 without Pd, a second
peak appeared at 720 K that also can be attributed to the burning
off of methyl groups. This low-temperature peak was not observed
for the HMDS-silylated Pd/SiO2 catalyst, due to the 4-h calcination
at 673 K of this catalyst before the silylation step. This calcination
removed the most reactive silanol groups (probably the vicinal and
geminal ones), leaving mostly isolated silanol groups. As a result,
on silylation, these samples formed only those methyl groups that
will burn off at 830 K, and the peak at 720 K was absent.
ing 0.06 mol/L of 4-IBAP, 0.12 mol/L of 4-IBPE, and 0.12 mol/L of
-IBEB, obtained from a previous hydrogenation experiment that
4
had been stopped at an appropriate point. The HMDS-silylated sil-
ica turned out to be catalytically inactive in both experiments.
During the hydrogenation experiments, liquid samples were ob-
tained at different times and analyzed with a Chrompack gas chro-
matograph (CP9001 autosampler), equipped with a CP Sil 8 CB
column (50 m × 0.25 mm), under the following conditions: FID
temperature, 523 K; injector temperature, 523 K; and column tem-
perature ramped uniformly from 423 to 523 K at 6.7 K/min.
2
.4. Adsorption experiments
Adsorption studies were performed by suspending 0.05 g of cat-
alyst in 1.1 mL of reaction mixture in a closed 25-mL flask under
stirring. The reaction mixture was obtained from a hydrogena-
tion experiment that had been stopped at the appropriate point to
achieve approximately the same concentration for 4-IBAP, 4-IBPE,
and 4-IBEB. An additional amount of 4-IBAP had to be added to
the reaction mixture to achieve ca. 0.12 mol/L of 4-IBAP, 4-IBPE,
and 4-IBEB. The adsorption experiments were carried out at con-
stant temperature of 293 K in a shaker (MS1 Minishaker, IKA)
at 800 rpm. Samples were obtained after sufficient equilibration
time (4 h) and analyzed by GC as described above, to measure
the liquid-phase equilibrium concentrations. The amounts of the
competitively adsorbed components were calculated from the mass
balance.
In the spent catalysts, the different peaks can be assigned to
(
a) molecular water physisorbed on the support (T < 380 K; en-
dothermal); (b) hydrocarbons physisorbed on the support (T =
380–530 K; endothermal), mainly 4-IBAP, as can be deduced by
comparing the profiles of the SiO2-HMDS sample exposed only to
a solution of 4-IBAP and of the sample SiO2-HMDS sample ex-
posed mainly to 4-IBPE and 4-IBEB; (c) hydrocarbons chemisorbed
on the Pd (T = 510–580 K) (profiles of the SiO2-HMDS sam-
ples did not exhibit this exothermal peak); (d) hydrocarbons
chemisorbed on the support (T = 580–680 K; exothermal), mainly
4-IBPE and 4-IBEB; and (e) methyl groups bound to the support
(
T = 750–850 K; exothermal).
3
. Results
In the spent catalysts, the broad peak at 465 K corresponding
to 4-IBAP physisorbed on the silica was almost insignificant for
the HMS and DDMS-silylated catalysts and negligible for the most
hydrophobic HMDS-silylated catalysts. In the transient TGA exper-
iment, this endothermal desorption cooled the catalyst, which had
such a significant effect in the unsilylated catalyst so as to post-
pone the combustion of the material chemisorbed on the Pd to
580 K. To prove that this shift indeed was due solely to the tran-
sient in the TGA, we performed a separate control experiment at
a much lower heating rate (2 K/min), and found that the peak for
3
.1. Characterization of the catalysts
Table 1 gives the Pd loading (determined by ICP) and the aver-
age metal crystal size (determined by TEM) of all catalysts pre-
pared. Pd was well dispersed in all of the fresh catalysts, with
particle sizes of 2–4 nm. TEM analyses were performed for Pd/SiO2
catalyst in each stage (fresh, reduced, used, and regenerated cat-
alyst) to study the size evolution during a complete use cycle.
Table 1 also reports textural properties.

A. Quintanilla et al. / Journal of Catalysis 257 (2008) 55–63
59
Fig. 3. Differential thermogravimetric profiles in air of the fresh (a) and spent (b)
Pd/SiO2 catalysts as prepared and after different silylation reactions (10 K/min).
material chemisorbed on the Pd occurred at 40 K lower than in
the 10 K/min experiment.
The fresh CAT-HMDS exhibited a small exothermal peak at
around 470 K, ascribed to material chemisorbed on the Pd, because
this peak was not observed in the HMDS-silylated SiO2 without Pd.
Fig. 4 shows the products that came off of the fresh HMDS-
silylated catalysts on heating, as determined by TPO-MS. The five
profiles correspond to the evolution of NO, NO2, CO, CO2, and H2O.
The figure shows that not only methyl groups, but also nitrogen-
containing groups (amines) were incorporated on the Pd during
silylation of the samples. Because amines are known to poison Pd
sites, leading to loss of activity [23], the CAT-HMDS catalyst was
thermally treated at 520 K for 4 h in a static air oven to remove
these amines as well as the methyl groups. The resulting catalyst,
designated CAT-HMDS520, did not exhibit a peak at 470 K in the
DTG profile (Fig. 3), although some remaining methyl and amine
groups would be expected to remain, according to the TPO-MS
profiles shown in Fig. 4.
Fig. 4. TPO-MS profiles of the fresh HMDS-silylated catalysts before and after heat
treatment at 520 K in a static air.
some methyl groups were bonded to the silica, as indicated by the
exothermal peak at 825 K in the DTG profile (Fig. 3) and by the
−1
bands at 2963 and 2908 cm
in the DRIFT spectra (Fig. 5), corre-
sponding to the stretching vibration of C–H bonds in CH3 and CH2
groups, respectively. Only with CAT-HMDS were new bands at 3366
To verify that the methyl groups were grafted onto the sil-
ica by silylation according to Fig. 2, DRIFT spectra were evaluated
−1
and 3277 cm
detected corresponding to primary and secondary
amine groups, respectively, corroborating the TGA and TPO-MS re-
sults.
(
Fig. 5). The presence of CH3Si was confirmed by the new band
−1
at 1260 cm . All of the silylated catalysts except CAT-HMS ex-
hibited this band. Moreover, the absorption band at 3740 cm
−1
,
well visible for Pd/SiO2, which can be attributed to the presence
of isolated or geminal silanol groups, disappeared in all of the sily-
lated catalysts (although a weak shoulder remained in CAT-HMS).
Although the extent of silylation with HMS was limited, as can
be deduced from the absence of methyl silicon bonds (Fig. 5),
3.2. Activity, stability, selectivity, and adsorption measurements
Fig. 6a shows a typical concentration–time profile of the vari-
ous components involved in the hydrogenation of 4-IBAP using a
fresh, nonsilylated Pd/SiO2 catalyst. These results confirm the pro-

60
A. Quintanilla et al. / Journal of Catalysis 257 (2008) 55–63
Fig. 5. DRIFT spectra of the fresh untreated and silylated catalysts and spent untreated Pd/SiO2 recorded at 323 K in 20% O2 in He.
posed reaction pathway shown in Fig. 1. The same three species
were always detected in hydrogenation with the different silylated
catalysts; no other components were found in the GC analysis.
Fig. 6b shows the overall selectivity (SIBPE) and yield (YIBPE) for
the alcohol as a function of 4-IBAP conversion (XIBAP). These vari-
ables were calculated as follows:
CIBPE
SIBPE =
(11)
(12)
(13)
0
C
− CIBAP
IBAP
YIBPE = XIBAP · SIBPE
0
C
− CIBAP
IBAP
XIBAP =
0
C
IBAP
According to the results, a selectivity up to 80% and a maxi-
mum yield around 60% could be expected at 373 K and 20 atm of
hydrogen pressure for the nonsilylated Pd/SiO2 catalyst.
Fig. 7 shows the effect of silylation on initial activity (i.e., the
consumption rate of 4-IBAP at minimal conversion levels, per kg of
Pd) at various moments in a deactivation–regeneration cycle. Ac-
tivity is reported for fresh catalysts, for spent catalysts reused in
a consecutive run in which fresh reactant is added to the con-
verted reaction mixture from the first run, and for a second run
after catalyst regeneration. The reproducibility, indicated by the er-
ror bars in Fig. 7, was calculated as the standard deviation from
repeated experiments. For the nonsilylated catalysts, wide ranges
of selectivity values (40% to 80%) and yield (35% to 60%) were
found.
The initial activity and the stability of Pd/SiO2 catalyst were
influenced by the silylating agent. Silylation with DDMS resulted
in the highest activity and stability; in contrast, the HMS-silylated
catalyst was completely deactivated after the first run. The effect
of regeneration in terms of activity recovery was excellent for the
untreated Pd/SiO2 catalyst (with TGA and DRIFT spectra demon-
strating the complete removal of the physisorbed and chemisorbed
hydrocarbons and coke); good for CAT-HMDS, CAT-HMDS520 and
CAT-DDMS; and detrimental for CAT-HMS.
Fig. 6. Relative concentrations as a function of reaction time (a) and overall se-
lectivity and yield of 4-IBPE as a function of 4-IBAP conversion (b) during hy-
drogenation of 4-IBAP with Pd/SiO2 catalysts under standard reaction conditions:
CIBAP = 0.27 mol/L, T = 373 K, pH2 = 20 bar, wcat = 0.1 g, solvent n-decane, stir-
ring rate = 1500 rpm

A. Quintanilla et al. / Journal of Catalysis 257 (2008) 55–63
61
Fig. 7. Effect of silylating agents on the initial activity of Pd/SiO2 catalyst for
Fig. 9. Amounts adsorbed at room temperature from a mixture of equal amounts
of reactants and products. Error bars estimated by the error propagation analysis.
Conditions: CIBAP ≈ 0.12 mol/L, CIBPE ≈ 0.12 mol/L, CIBEB ≈ 0.12 mol/L, T = 293 K,
p = 1 bar, t = 4 h, wcat = 0.05 g, VL = 1.1 mL, solvent n-decane.
4
-IBAP hydrogenation under standard conditions: CIBAP = 0.27 mol/L, T = 373 K,
pH = 20 bar, wcat = 0.1 g, solvent n-decane, stirring rate = 1500 rpm. Legend:
2
(
) 1st run; (1) 2nd run after injection of additional reactant to the mixture of the
st run; (2) 2nd run after regeneration. Error bars indicate the standard deviation
1
from the repeated experiments.
selectivity differed between the fresh samples (top plot) and the
samples after deactivation and regeneration (bottom plot). Selec-
tivity was clearly affected by the silylating agent. Only the HMDS-
silylated catalysts exhibited greater selectivity than the nonsily-
lated catalysts. The selectivity of the DDMS catalyst dropped to
60%. The hardly active HMS-silylated catalyst exhibited a peculiar
behavior, with its (relatively low) selectivity increasing during the
conversion of 4-IBAP. After regeneration, the untreated Pd/SiO2 cat-
alyst lost most of its original selectivity, whereas the appropriate
silylated catalysts remained selective, in a range of 70%–80%.
The results of the competitive adsorption experiments, collected
in Fig. 9, show that in general, 4-IBPE adsorbed more favorably
than 4-IBAP on the fresh and used catalysts. The effect of silylation
and regeneration in relation to the catalyst selectivity is discussed
in the next section.
4
. Discussion
4
.1. Silylation
The effectiveness of the silylation quantified from the weight
loss in the TGA experiments can be ranked as follows: CAT-HMDS
CAT-HMDS520 (10%) > CAT-HMS (8.5%) > CAT-DDMS (6%). This
=
may be related to the stoichiometry of the silylation reactions.
With DDMS, for each silanol group, one methyl group was incor-
porated, compared with three methyl groups with HMDS and HMS
(
Fig. 2). These values are in the same range as those reported in
previous studies using the same HMDS:sample ratio [1,3].
Silylation modified the support and did not lead to significant
coverage of the Pd, except for HMDS. Only for HMDS did the char-
acterization techniques (TGA, DRIFT and TPO) indicate the blocking
of active sites by amine groups (Figs. 3, 4, and 5) and fouling by
carbon groups (Figs. 3 and 5).
The effectiveness of treatment in terms of hydrophobicity, as
reflected by the amount of water desorbed below 380 K (Fig. 3),
can be ranked as CAT-HMDS > CAT-HMDS520 > CAT-DDMS >
CAT-HMS = Pd/SiO2. The HMS silylating agent seemed to have no
significant affect on the hydrophobic–hydrophilic properties of the
Pd/SiO2; as deduced earlier based on the DRIFT spectra (Fig. 5), the
yield of the silylation was quite limited. Based on these findings,
we can conclude that silylating agents that contain a functional
group or a heteroatom (i.e., amine or halogen) are more effective.
Fig. 8. Effect of silylating agent on the overall selectivity to 4-IBPE as a function
of 4-IBAP conversion of fresh (a) and regenerated (b) Pd/SiO2 catalysts; results ob-
tained under standard conditions: CIBAP = 0.27 mol/L, T = 373 K, pH2 = 20 bar,
wcat = 0.1 g, solvent n-decane, stirring rate = 1500 rpm. Legend: (1) Pd/SiO2 cata-
lyst; (") CAT-HMDS; (!) CAT-HMDS520; (E) CAT-HMS; and (Q) CAT-DDMS.
Whereas Fig. 7 shows only activity data, Fig. 8 shows the evo-
lution of selectivity toward the alcohol during conversion for dif-
ferently silylated samples and untreated Pd/SiO2. The evolution of

62
A. Quintanilla et al. / Journal of Catalysis 257 (2008) 55–63
HMDS was the most efficient silylating agent, providing the
most hydrophobic Pd/SiO2 surface. It is also the most stable among
those investigated (Fig. 3). Comparing the DTG profiles of the fresh
and used silylated catalysts (Fig. 3), we can deduce that the HMDS-
silylated catalyst lost only a small fraction of the methyl groups
during the hydrogenation reaction.
DDMS did not act as a stable silylating agent for the hydrogena-
tion of aromatic ketones. The DTG results showed a complete ab-
sence of methyl groups after the reaction. Moreover, in the DRIFT
spectra of the used DDMS-silylated catalyst (not shown), a new
−1
band appeared at 3740 cm , attributed to terminal or geminal
silanol groups. This can be explained by the interaction between
Si–CH3 groups and water produced on the hydrogenation reaction,
resulting in hydrolysis of the O–Si bonds. This structure is known
to be less stable in the presence of water [24].
◦
During the in situ reduction (at 100 C and 10 atm) of Pd/SiO2,
the dispersion decreased by 50% and it decreased no further dur-
ing the reaction (Table 1). It is significant that this did not affect
the activity; the activity of the Pd/SiO2 catalyst was fully recovered
after regeneration (Fig. 7).
Fig. 10. Correlation of the relative adsorption strength of 4-IBPE to 4-IBAP with the
selectivity at 30% of 4-IBAP conversion. Conditions: CIBAP ≈ 0.12 mol/L, CIBPE ≈
0
.12 mol/L, CIBEB ≈ 0.12 mol/L, T = 293 K, p = 1 bar, t = 4 h, wcat = 0.05 g, VL =
.1 mL, solvent n-decane.
Silylation did not promote Pd sintering. Spent silylated cata-
lysts exhibited a similar decrease in dispersion as the untreated
catalysts except when amines and methyl groups were poisoning
the Pd sites, as in the HMDS-silylated catalyst, in which case Pd
sintering seemed to be suppressed (Table 1). Only silylation with
HMDS, with no post-treatment, affected the textural properties of
the Pd/SiO2 catalyst. The HMDS-silylated catalyst had a slightly re-
duced specific surface area and pore volume, due to the presence
of alkylsilane groups at the surface of the pores, in agreement with
previous reports [3,7].
1
relative adsorption of 4-IBPE/4-IBAP on the initial overall selectiv-
ity. The regenerated Pd/SiO2 catalyst demonstrated a high relative
adsorption 4-IBPE/4-IBAP value compared with the fresh catalyst,
but with a decrease in selectivity.
The silylating agent had a strong impact on the selectivity of
fresh and regenerated catalysts. Only HMDS, the most effective
silylating agent, improved the selectivity of the untreated catalyst.
The most remarkable finding is that the properly silylated catalysts
Silylation did not provoke Pd leaching under the conditions
used here in any of the agents except DDMS. In this case, to avoid
strong leaching of Pd (ca. 70%), the silica support was first sily-
lated, followed by Pd application. As a result, the Pd loading was
lower than the target value due to the smaller amount of anchor-
ing centers for Pd in the silica surface.
(
HMDS and DDMS) remained selective after regeneration (Fig. 8).
This can be explained in terms of the adsorption strength of re-
actants and products on the support, not on the Pd sites (Figs. 3
and 10). In these regenerated silylated catalysts, the adsorption of
4
-IBPE relative to 4-IBAP did not change as much (CAT-HMDS) or
even decreased (CAT-DDMS) compared with their fresh counter-
parts. Their behavior on regeneration was even more spectacular,
with selectivity remaining at ca. 80% (Fig. 10).
4
.2. Selective hydrogenation
Based on these findings, we can conclude that the competi-
tive adsorption of the reactant and product on the catalyst had an
important effect on the selectivity of this hydrogenation reaction,
a correlation illustrated in Fig. 10. Silylation reduced the polarity
of the surface and thus the adsorption of ketone 4-IBAP and al-
cohol product 4-IBPE. The likelihood of alcohol forming hydrogen
bridges with the surface was reduced significantly, so that the rel-
ative composition of the adsorbed phase on the metal changed
in favor of the ketone by equilibria such as shown in Eq. (10).
This increased the first hydrogenation step [Eq. (4)] relative to the
undesired second hydrogenation [Eq. (7)], resulting in a more fa-
vorable yield of alcohol. Apparently, silylation also stabilized the
relative adsorption of the product 4-IBPE to the reactant 4-IBAP on
regeneration of the catalyst, in contrast with the situation for the
untreated Pd/SiO2 catalyst, in which this changed unfavorably on
regeneration (Fig. 9), promoting the consecutive reaction and re-
sulting in low selectivity.
Previously, silylation to reduce the surface polarity of catalysts
has been applied mainly to selective oxidation reactions, for which
the products obviously are more polar than the reactants and are
more prone to consecutive conversion or strong adsorption. The
present study has demonstrated that this approach can be ap-
plied to certain selective hydrogenations as well, when the prod-
ucts formed interact through hydrogen bridges with the surface
silanol groups, such as in the hydrogenation of a ketone, alde-
hyde, or epoxide to an alcohol. Clearly, in such cases, changing the
hydrophobic properties of catalyst surfaces can improve reaction
selectivity.
Silylation affected the hydrophobic character of the catalysts
and, consequently, the adsorption of organic reactants and prod-
ucts (Fig. 9) and thus the activity (Fig. 7) and selectivity (Fig. 8).
All silylated catalysts would be expected to be less active than the
untreated catalyst, because the Pd active sites are exposed to lower
local concentrations of 4-IBAP (Fig. 9). Our experiments shown in
Fig. 7 confirm this hypothesis, except for the fresh DDMS-silylated
catalyst. This higher activity may be due to a promoting effect of
chlorides remaining on the support after silylation. The low activity
seen in the HMDS-silylated samples can be attributed to poisoning
of the Pd sites by amine groups, as confirmed by comparing the
activity of CAT-HMDS and the heat-treated sample CAT-HMDS520,
in the latter, the fresh catalyst demonstrated greater activity.
The decreased activity in the nonmodified Pd/SiO2 catalyst
from 25 to 1 mol/(kgPd s) in a consecutive run can be attributed
to the large amount of 4-IBAP physisorbed/chemisorbed on the sil-
ica surface and chemisorbed on the Pd sites (Fig. 3). In agreement
with this, DRIFT spectra (Fig. 5) of the untreated Pd/SiO2 catalyst
confirmed the loss of isolated hydroxyl groups on the silica and
the appearance of methyl groups, alkoxy groups, and silyl ether
groups. Hydration of the catalyst also occurred, as indicated by the
−1
increased band range from 3500 to 3400 cm . Thus, Pd/SiO2 cat-
alyst deactivated during the reaction ([16], Fig. 7) due to surface
deposition mainly on the support and also on the Pd.
Although regeneration removed the deposits and allowed the
recovery of activity, it changed the adsorption properties of un-
treated Pd/SiO2 catalyst (Fig. 9). Fig. 10 shows the effect of the

A. Quintanilla et al. / Journal of Catalysis 257 (2008) 55–63
63
5
. Conclusion
MS, and DRIFTS setups and Sander Brouwer for the N2 adsorption/
desorption isotherm measurements.
In this study, the selectivity of Pd/SiO2 catalysts in the hydro-
genation reaction of an aromatic ketone to the corresponding al-
cohol was significantly improved by silylating the catalyst. A clear
correlation between the suppression of adsorption of the desired
aromatic alcohol and the suppression of the consecutive reaction
of that alcohol was observed. Silylation was found to decrease
the relative adsorption strength of the desired alcohol to the ke-
tone reactant, suppressing further hydrogenation of the alcohol
and maintaining the selectivity. Silylation was especially effective
for the regenerated catalyst in which the untreated Pd/SiO2 had
lost most of its original selectivity.
HMDS proved to be the most effective silylation agent among
those investigated in terms of incorporation of methyl groups, cat-
alyst hydrophobicity, and stability. The use of a silylating agent
with a functional group or a heteroatom was the most effective
approach for obtaining appropriate silylation yields. Because silyla-
tion does not cover active Pd, as long as the silylating agent does
not cause the metal leaching, silylation of the catalyst is preferred
over silylation of the support, which leads to a lower metal load-
ing due to removal of the isolated hydroxyl groups where Pd is
anchored.
References
[1] A. Corma, M. Domine, J.A. Gaona, J.L. Jorda, M.T. Navarro, F. Rey, J. Perez-
Pariente, J. Tsuji, B. McCulloch, L.T. Nemeth, Chem. Commun. (20) (1998)
2211.
[2] T. Tatsumi, K.A. Koyano, N. Igarashi, Chem. Commun. (3) (1998) 325.
[3] M.V. Cagnoli, S.G. Casuscelli, A.M. Alvarez, J.F. Bengoa, N.G. Gallegos, M.E. Criv-
ello, E.R. Herrero, S.G. Marchetti, Catal. Today 107–108 (2005) 397.
[4] F. Bauer, W.H. Chen, H. Ernst, S.J. Huang, A. Freyer, S.B. Liu, Microporous Meso-
porous Mater. 72 (1–3) (2004) 81.
[
5] A.J. van Roosmalen, M.C.G. Hartmann, J.C. Mol, J. Catal. 66 (1) (1980) 112.
[6] S.R. Zheng, A. Jentys, J.A. Lercher, J. Catal. 241 (2) (2006) 304.
[7] M. Ojeda, F.J. Perez-Alonso, P. Terreros, S. Rojas, T. Herranz, M.L. Granados, J.L.G.
Fierro, Langmuir 22 (7) (2006) 3131.
[8] X. Li, H. Gao, G. Jin, L. Chen, L. Ding, H. Yang, Q. Chen, J. Mol. Struc. 872 (2007)
1
0.
[9] R.R. Sever, R. Alcala, J.A. Dumesic, T.W. Root, Microporous Mesoporous
Mater. 66 (2003) 53.
[10] N. Igarashi, K. Hashimoto, T. Tatsumi, Microporous Mesoporous Mater. 104
(
2007) 269.
11] K. Lin, L. Wang, F. Meng, Z. Sun, Q. Yang, Y. Cui, D. Jiang, F. Xiao, J. Catal. 235
2005) 423.
[12] A. Chica, A. Corma, M.E. Domine, J. Catal. 242 (2006) 299.
[
(
[
[
[
[
13] V.A. Veefkind, J.A. Lercher, J. Catal. 180 (1998) 258.
14] F. Zaccheria, N. Ravasio, M. Ercoli, P. Allegrini, Tetrahedron Lett. 46 (45) (2005).
15] V. Elango, EP 0400 892 A2 (1990), to Hoechst Celanese Corp.
16] N. Thakar, W. Buijs, F. Kapteijn, J.A. Moulijn, J. Catal. 248 (2) (2007) 249.
Acknowledgments
[17] N. Thakar, K. Djanashvili, H. van Bekkum, F. Kapteijn, J.A. Moulijn, J. Catal.
2
46 (2) (2007) 344.
Financial support was provided by the Spanish Ministry of
Education and Science through postdoctoral fellowship EX-2006-
[
[
18] N. Thakar, R.J. Berger, f. Kapteijn, J.A. Moulijn, J. Catal. 62 (2007) 5322.
19] M.B.J. Roeffaers, B.F. Sels, D. Loos, C. Kohl, K. Mullen, P.A. Jacobs, J. Hofkens, D.E.
De Vos, ChemPhysChem 6 (11) (2005) 2295.
1012. The Dutch National Center for High Resolution Electron
Microscopy, particularly Drs. Valerie C. L. Butselaar and Patricia
J. Kooyman of Delft University of Technology, are acknowledged
for performing the electron microscopy investigations. Professor
W. Buijs is gratefully thanked for fruitful discussions. The authors
also thank Bart van der Linden for his assistance with TGA, TPO-
[20] A.P. Rao, G.M. Pajonk, A.V. Rao, J. Mater. Sci. 40 (13) (2005) 3481.
21] F. Omota, A.C. Dimian, A. Bliek, Appl. Catal. A Gen. 294 (2) (2005) 121.
22] E. Shirakawa, K. Hironaka, H. Otsuka, T. Hayashi, Chem. Commun. (37) (2006)
927.
23] H. Sajiki, K. Hattori, K. Hirota, J. Chem. Soc. Perkin Trans. 1 (24) (1998) 13981.
[
[
3
[
[24] A.P. Rao, A.V. Rao, G.M. Pajonk, Appl. Surf. Sci. 253 (2007) 6032.